Ubiquitination-mediated Regulation of Biosynthesis of the Adhesion Receptor SHPS-1 in Response to Endoplasmic Reticulum Stress*

Misfolding of proteins during endoplasmic reticulum (ER) stress results in the formation of cytotoxic aggregates. The ER-associated degradation pathway counter-acts such aggregation through the elimination of misfolded proteins by the ubiquitin-proteasome system. We now show that SHP substrate-1 (SHPS-1), a transmembrane glycoprotein that regulates cytoskeletal reorgani-zation and cell-cell communication, is a physiological substrate for the Skp1-Cullin1-NFB42-Rbx1 (SCF NFB42 ) E3 ubiquitin ligase, a proposed mediator of ER-associ-ated degradation. SCF NFB42 mediated the polyubiquitination of immature SHPS-1 and its degradation by the proteasome. Ectopic expression of NFB42 both sup-pressed the formation of aggresome-like

Adhesion molecules on the surface of cells play pivotal roles in many physiological processes, including cell growth, differentiation, and migration. Dysregulation of specific adhesion molecules has thus been shown to result in pathological conditions such as inflammation, neurodegeneration, and cancer metastasis. We previously identified SHP substrate-1 (SHPS-1) 1 (1,2), an adhesion receptor of the immunoglobulin superfamily also known as SIRP␣1 (3), BIT (4), MFR (5), and p84 neural adhesion molecule (6). The extracellular region of SHPS-1 comprises three immunoglobulin-like domains, the most amino-terminal of which associates with the ligand CD47 (7). Through its interaction with CD47, SHPS-1 contributes to the phagocytosis of red blood cells by macrophages (8), to macrophage multinucleation (9), to T cell activation (10), and to neutrophil transmigration (11).
The cytoplasmic region of SHPS-1 acts as a scaffold for the assembly of multiprotein complexes (1,3,12). By recruiting the protein-tyrosine phosphatase SHP-2 to the cell membrane, SHPS-1 negatively or positively regulates intracellular signaling initiated either by tyrosine kinase-coupled receptors for growth factors or by cell adhesion to extracellular matrix proteins (13,14). Our studies with mice lacking most of the cytoplasmic region of SHPS-1 have also revealed that this protein is an upstream signaling component important for SHP-2-mediated regulation of cell migration and cytoskeletal reorganization (15,16).
Expression of SHPS-1 in the central nervous system is regulated both during embryonic development and postnatally, and this regulation has been implicated in synapse formation or maintenance (6,17). SHPS-1 is also down-regulated during cell transformation and thus might play a role in carcinogenesis (14). Little is known, however, of the mechanisms by which the cellular abundance of SHPS-1 is determined, although several oncogene products down-regulate SHPS-1 mRNA (18).
In the present study, we sought to provide insight into SHPS-1 function by identifying proteins that regulate SHPS-1 or transduce signals emanating from its cytoplasmic region. We now demonstrate a physical and functional interaction between SHPS-1 and both neural F box protein of 42 kDa (NFB42) and S phase kinase-associated protein 1 (Skp1) in mouse brain and melanocytes. NFB42 (recently renamed Fbx2) was first identified as a gene product that is highly enriched in the nervous system (19). F box proteins contain an F box domain and mediate substrate recognition by Skp1-Cullin1-F box protein-Rbx1 (SCF)-type E3 ubiquitin-protein ligase complexes (20). Ectopic expression of NFB42 was shown to inhibit the proliferation of neuroblastoma cells (19). Skp1 is an invariant component of SCF-type E3 enzymes and is thought to control the cell cycle (21). NFB42 and Skp1 are thus implicated in the control of neuronal proliferation by protein ubiquitination, although the biological roles of these proteins remain to be established. Skp1 also participates in various signal transduction pathways, including that responsible for glucose-dependent reassembly of V-ATPase complexes (22).
We now show that NFB42 and Skp1, together with Cullin1, constitute an SCF-type E3 ubiquitin ligase (SCF NFB42 ) and that this complex catalyzes the polyubiquitination of immature forms of SHPS-1. Forced expression of NFB42 resulted in elimination of misfolded SHPS-1 molecules from the endoplasmic reticulum (ER) by the ubiquitin-proteasome proteolytic pathway, an effect that was associated with marked inhibition of the formation of cellular aggregates containing SHPS-1. Furthermore, this elimination of misfolded SHPS-1 from the ER led to substantial up-regulation of SHPS-1 expression at the cell surface. We also provide evidence that SCF NFB42 functions to maintain ER homeostasis during cellular stress by supporting the biosynthesis of SHPS-1.

EXPERIMENTAL PROCEDURES
Expression Vectors-The cDNAs encoding the SHPS-1 mutants SHPS-1-4F, in which all four tyrosine residues in the cytoplasmic region (Tyr 408 , Tyr 432 , Tyr 449 , and Tyr 473 ) are replaced by phenylalanine, and SHPS-1-JM, in which two juxtamembrane lysine residues (Lys 401 and Lys 402 ) are replaced by asparagine and glutamine, respectively, were generated by site-directed mutagenesis with the full-length mouse SHPS-1 cDNA (2) as a template and a Transformer Site-directed Mutagenesis kit (Clontech). The wild-type and mutant SHPS-1 cDNAs were then inserted separately into the EcoRI and NotI site of pTracer-CMV (Invitrogen). To generate a cDNA encoding SHPS-1 fused to green fluorescent protein (GFP), we performed the polymerase chain reaction (PCR) with a full-length human SHPS-1 cDNA (2) as a template, a T7 primer, and the antisense primer 5Ј-AAAGTCGACTTCTTCTACAAGG-3Ј. The PCR product was digested with EcoRI and SalI and then inserted into pEGFP N2 (Clontech). The pSR␣ vector encoding wildtype rat SHPS-1 was described previously (1). The pcDNA3 vectors encoding hemagglutinin epitope (HA)-tagged versions of either wildtype rat NFB42 (amino acids 2-296) or the deletion mutants NFB⌬P (residues 53-296), NFB⌬P⌬F (residues 95-296), or NFBPF (residues 2-94) (19) were kindly provided by R. Pittman (University of Pennsylvania). To generate cDNAs encoding Myc epitope-tagged versions of wild-type NFB42 and NFB⌬P⌬F, we amplified the corresponding coding regions by PCR with the full-length NFB42 cDNA as template. The PCR products were digested with EcoRI and SalI and then inserted in-frame into pCl-neo (Invitrogen) that had been modified to add the coding sequence for the Myc epitope to the 5Ј end of the inserted cDNA. The pcDNA3 vectors encoding Myc epitope-tagged forms of mouse Skp1 and mouse Cullin1 were kindly provided by S. Hatakeyama (Kyushu University, Japan).
A rat monoclonal antibody (mAb) to mouse SHPS-1 (24) and rabbit polyclonal antibodies to NFB42 (19) were described previously. Mouse mAbs to human SIRP␣1 and to Skp1 were obtained from Transduction Laboratories; rabbit polyclonal antibodies to SHPS-1 and to Ser 51phosphorylated eukaryotic initiation factor 2␣ (eIF2␣) were from Upstate Biotechnology, Inc.; rabbit polyclonal antibodies to eIF2␣ were from Cell Signaling Technology; rabbit polyclonal antibodies to Cullin1 were from Zymed Laboratories Inc.; a mouse mAb to ubiquitin (P4D1) and normal rat or mouse IgG were from Santa Cruz Biotechnology; Texas Red-conjugated sheep polyclonal antibodies to mouse IgG were from Amersham Biosciences; and Alexa Fluor 488-conjugated goat polyclonal antibodies to rat IgG were from Molecular Probes. The mAb 9E10 to the Myc tag and mAb 12CA5 to the HA tag were purified from the culture supernatants of mouse hybridoma cells.
Purification of SHPS-1-binding Proteins-All procedures were performed at 4°C. The entire brain from one male mouse (12 weeks of age) was disrupted in 10 ml of ice-cold buffer A (10 mM Tris-HCl (pH 7.6), 150 mM NaCl, 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, aprotinin (100 units/ml), 1 mM benzamidine, 10 M leupeptin, 1 mM sodium vanadate) with a Polytron homogenizer. The crude extract was centrifuged at 100,000 ϫ g for 60 min, and the resulting supernatant (protein, 10 mg/ml) was incubated for 3 h in a 15-ml tube on a gently rotating platform with protein G-Sepharose beads (500 l of beads) (Amersham Biosciences) that had been covalently coupled with 500 g of the mAb to mouse SHPS-1. The beads were then washed four times with buffer A and once with 10 mM Tris-HCl (pH 6.8), and bound proteins were eluted with 5 ml of acidic buffer containing 100 mM glycine (pH 2.5). Fractions (1 ml each) were collected, neutralized by the addition of 0.1 volume of 1 M Tris-HCl (pH 8.8), and subjected to dot-immunoblot analysis with polyclonal antibodies to SHPS-1. Two peak fractions containing SHPS-1 were combined and concentrated with a Centricon 30 (Amicon) filtration device to a final volume of 50 l.
Mass Spectrometry-Proteins were resolved by SDS-PAGE on a 12.5% gel and stained with silver. Bands of interest were excised from the gel, washed in a solution comprising acetonitrile and 100 mM NH 4 HCO 3 (1:1, v/v), and subjected to in-gel digestion with trypsin overnight at 37°C. Peptides were then extracted with a solution of 50% acetonitrile and 5% formic acid, lyophilized, resuspended in 0.1% acetic acid, and analyzed with an LCQ ion-trap mass spectrometer equipped with a nanospray ion source and an HP1100 liquid chromatography system (ThermoQuest). Peptide sequences were identified by searching the NCBI nonredundant sequence data base with the obtained tandem mass spectra with the use of Mascot software (Matrix Science).
Immunoprecipitation, Lectin Binding, and Immunoblot Analysis-Cells were lysed on ice in lysis buffer (10 mM Tris-HCl (pH 7.6), 150 mM NaCl, 1% Triton X-100) containing 1 mM phenylmethylsulfonyl fluoride, aprotinin (100 units/ml), and 1 mM sodium vanadate. The lysates were centrifuged at 10,000 ϫ g for 15 min at 4°C. For immunoprecipitation, the supernatants (protein, ϳ5 mg/ml) were incubated for 3 h at 4°C with antibody-coupled protein G-Sepharose beads. Alternatively, the supernatants were incubated for 1 h at 4°C in the presence of 1 mM MnCl 2 and 1 mM MgCl 2 with agarose beads conjugated with Galanthus nivalis agglutinin (EY Laboratories). Both types of beads were then washed four times with lysis buffer, suspended in Laemmli sample buffer, and subjected to SDS-PAGE and immunoblot analysis with various antibodies and the ECL detection system (Amersham Biosciences).
Immunofluorescence Analysis-All procedures were performed at room temperature. Cells seeded on glass coverslips were washed with phosphate-buffered saline (PBS), fixed with 3% paraformaldehyde in PBS for 20 min, and incubated with 50 mM NH 4 Cl for 10 min. Cells were then permeabilized for 2 min in PBS containing 0.1% Triton X-100 and 1% bovine serum albumin (BSA) before incubation first for 1 h in PBS containing 0.1% Triton X-100, 10% FBS, and 0.5% BSA and then for 1 h with mAb 9E10 to the Myc tag. After washing twice with PBS containing 0.5% BSA, the cells were incubated with Texas Red-conjugated secondary antibodies for 30 min, washed three times with PBS containing 0.5% BSA, and examined with a laser-scanning confocal microscope (Bio-Rad model MRC-1024).
Flow Cytometry-Detached cells (0.5 ϫ 10 6 to 1 ϫ 10 6 ) were washed with buffer B (PBS containing 1 mM EDTA and 2% FBS) and then incubated with the mAb to mouse SHPS-1 (20 g/ml) or with control rat IgG (20 g/ml) for 1 h on ice. They were washed with buffer B and incubated for 1 h on ice with Alexa Fluor 488-conjugated secondary antibodies (10 g/ml). The stained cells were washed twice, suspended in 1 ml of buffer B, and then analyzed with a FACSCalibur flow cytometer (BD Biosciences). Data were processed with CellQuest software (BD Biosciences).

SHPS-1 Forms a Complex with NFB42 and Skp1 in Vivo-
SHPS-1 is particularly abundant in the central nervous system (4,6,24). We therefore adopted an immunoaffinity approach to isolate proteins that interact with SHPS-1 in the brain. A detergent-solubilized extract of mouse brain was applied to protein G-Sepharose beads that had been coupled with a mAb that reacts with the extracellular portion of mouse SHPS-1 (24). After extensive washing of the beads, bound material was eluted, resolved by SDS-PAGE, and visualized by silver staining (Fig. 1A). Bands corresponding to polypeptides that copurified with SHPS-1 were excised from the gel and subjected to in-gel digestion with trypsin, and the resulting peptides were analyzed by tandem mass spectrometry. The peptide sequences so obtained were used to search the NCBI sequence data base with Mascot software. Data base searches revealed that peptide sequences derived from proteins of ϳ40 and ϳ20 kDa were identical to sequences of rat NFB42 (AF098301) (19) and mouse Skp1 (AF083214) (25), respectively (Fig. 1A). NFB42 and Skp1 were shown previously to interact directly through the F box FIG. 1. Identification of NFB42 and Skp1 as SHPS-1-binding proteins. A, mouse brain extract was subjected to immunoprecipitation (IP) with a mAb targeted to the extracellular portion of mouse SHPS-1 (␣SHPS-1) or with normal rat IgG (NRG), and the resulting precipitates were subjected to SDS-PAGE on a 12.5% gel and silver staining. Polypeptides excised from the gel were subjected to mass spectrometric analysis for identification. The positions of SHPS-1, NFB42, and Skp1, as well as those of molecular size standards (in kilodaltons), are indicated. Determined amino acid sequences corresponding to rat NFB42 or to mouse Skp1 are shown. B, mouse brain extract or Melan-a cell lysate was subjected to immunoprecipitation with the mAb to mouse SHPS-1, with a mAb targeted to the cytoplasmic region of human SIRP␣1, or with control rat or mouse (NMG) IgG. The immunoprecipitates were then subjected to immunoblot analysis with polyclonal antibodies to NFB42 (middle panel) or with a mAb to Skp1 (bottom panel). Duplicate precipitates were probed with polyclonal antibodies to SHPS-1 (top panel). Mouse brain extract and Melan-a cell lysate (Input) were also subjected directly to immunoblot analysis to determine the total amounts of each protein. C, mouse brain extract was subjected to immunoprecipitation with polyclonal antibodies to NFB42 (␣NFB42) or with control rabbit IgG. The immunoprecipitates domain of the former (19), suggesting that they associate with SHPS-1 as a complex.
Immunoblot analysis confirmed that NFB42 and Skp1 associate with SHPS-1 in mouse brain extract, although Skp1 was barely detectable in immunoprecipitates prepared with a mAb that reacts with the cytoplasmic region of human SIRP␣1 (Fig.  1B). Endogenous SHPS-1 was also coprecipitated with antibodies to NFB42 (Fig. 1C). Immunoblot analysis also revealed that an immortal line of mouse pigmented melanocytes, Melan-a (26), expresses SHPS-1, NFB42, and Skp1 at levels similar to those observed in mouse brain extract (Fig. 1B). A substantial proportion of endogenous NFB42 was associated with SHPS-1 in Melan-a cells, although this proportion was markedly less than that apparent in mouse brain. Skp1 interaction with SHPS-1 was not detected in these cells. The interaction between SHPS-1 and NFB42 was also demonstrated in CHO cells transiently expressing recombinant SHPS-1 and HA-tagged NFB42 (Fig. 1D); NFB42 associated preferentially with immature forms of SHPS-1 in these cells, hardly interacting at all with the mature forms (Fig. 1D).
Structural Requirements for SHPS-1-NFB42 Interaction-Immunoprecipitation and immunoblot analysis with a brain extract prepared from mice expressing an SHPS-1 mutant that lacks most of the cytoplasmic region (15) revealed that NFB42 and Skp1 each associated with the mutant protein, albeit to a lesser extent than with wild-type SHPS-1 ( Fig. 2A), suggesting that the extracellular or membrane-proximal cytoplasmic region of SHPS-1 mediates the interaction with NFB42. We next examined the ability to interact with NFB42 of SHPS-1 mutants in which two juxtamembrane lysine residues in the cytoplasmic region were replaced by asparagine and glutamine, respectively, or in which all four cytoplasmic tyrosine residues were changed to phenylalanine. Each of these mutant proteins associated with HA-tagged NFB42 in CHO cells to an extent similar to that observed with wild-type SHPS-1 (Fig. 2B). Thus, neither the positively charged juxtamembrane region nor cytoplasmic tyrosine residues of SHPS-1 appeared to play a major role in the interaction of this protein with NFB42.
To map the SHPS-1-binding site in NFB42, we employed the following three NFB42 mutant proteins: NFB⌬P, which lacks the NH 2 -terminal PEST domain; NFB⌬P⌬F, which lacks both the PEST and F box domains; and NFBPF, which comprises only the PEST and F box domains (Fig. 2C). The HA-tagged versions of these mutant proteins were transiently expressed in CHO cells that stably express wild-type mouse SHPS-1 (CHO-mSHPS-1 cells) and were then tested for their ability to interact with SHPS-1. NFB⌬P and NFB⌬P⌬F, but not NFBPF, bound to SHPS-1 (Fig. 2D), suggesting that the COOH-terminal domain of NFB42 mediates this interaction. Consistent with previous observations (19), NFB⌬P, but not NFB⌬P⌬F, also bound Skp1.
SCF NFB42 Ubiquitin Ligase Mediates Polyubiquitination and Proteasomal Degradation of Immature SHPS-1-SCF-type ubiquitin ligase (E3) complexes include the invariant component Skp1, an F box protein that is responsible for substrate recognition, and Cullin1, which serves as a scaffold for organization of the other subunits (27). Both NFB42 (Fig. 3A) and Skp1 (Fig. 3B) interact with Cullin1, suggesting that these three proteins constitute an SCF NFB42 ubiquitin ligase. Ectopic expression of NFB42 in CHO-mSHPS-1 cells revealed polyubiquitination of SHPS-1 in the presence of the specific proteasome inhibitor MG132 (Fig. 3C). Treatment of cells with MG132 also substantially increased the amount of immature SHPS-1 associated with NFB42 (Fig. 3D). Together, these results suggested that SCF NFB42 acts as an E3 that catalyzes the polyubiquitination of immature SHPS-1 and that the ubiquitinated protein is subsequently eliminated by proteasomal degradation.
SHPS-1 Is a Substrate for ER-associated Degradation-The possibility that SCF NFB42 -mediated degradation down-regulates the biological function of SHPS-1 appeared unlikely given the lack of a positive effect of MG132 on the abundance of mature SHPS-1 (Fig. 3D). Rather, the selective association of immature forms of SHPS-1 with NFB42 suggested a role for SCF NFB42 in the elimination of SHPS-1 molecules misfolded during biosynthesis. To test this possibility, we determined whether SHPS-1 serves as a substrate for ER-associated degradation (ERAD), a pathway by which misfolded proteins in the ER are translocated to the cytosol, polyubiquitinated, and degraded by the proteasome (28 -30). Fluorescence microscopy of Melan-a cells expressing a fusion construct of SHPS-1 and GFP (SHPS-1-GFP) revealed that the recombinant protein was localized predominantly to both the cell surface and the perinuclear region (Fig. 4A). A small proportion of SHPS-1-GFP was also detected as small aggregates throughout the cytoplasm. Similar analysis of cells exposed to MG132 showed that most SHPS-1-GFP was present either in large aggregates adjacent to the nucleus (Fig. 4B) or in small aggregates in the cytoplasm (data not shown). The large perinuclear aggregates exhibited the characteristic features of aggresomes, cellular inclusion bodies that consist of polyubiquitinated proteins and which are formed by ERAD substrates in response to inhibition of proteasome activity (31,32). Our results thus suggested that SHPS-1 is a substrate for the ERAD pathway.
SCF NFB42 Promotes Elimination of Misfolded SHPS-1 by the ERAD Pathway-Wild-type NFB42 was distributed diffusely in the cytosol of Melan-a cells incubated in the absence (data not shown) or presence (Fig. 4G) of MG132. Overexpression of wild-type NFB42 induced redistribution of a substantial proportion of SHPS-1-GFP to the cell surface and markedly suppressed the formation of aggresome-like bodies and cytoplasmic aggregates in cells treated with MG132 ( Fig. 4C; data not shown). In contrast, expression of NFB⌬P⌬F, which does not form a functional ubiquitin ligase, promoted the formation of these structures ( Fig. 4D; data not shown), indicative of a dominant negative effect of this mutant. The staining of wildtype NFB42 overlapped partially with SHPS-1-GFP fluorescence in a reticular pattern throughout the cytoplasm (Fig. 4I), consistent with colocalization of these proteins at or in the ER. However, extensive colocalization of the two proteins was not detected at the cell periphery, where mature SHPS-1 is expected to reside, even in the presence of MG132. A substantial proportion of NFB⌬P⌬F colocalized with SHPS-1-GFP in the aggresome-like bodies of MG132-treated cells (Fig. 4J).
Binding experiments in vitro with G. nivalis agglutinin lectin, which specifically recognizes terminal mannose residues of glycoproteins (33), revealed that recombinant SHPS-1 expressed in CHO cells exists in several distinct forms with regard to N-glycan structure (Fig. 5A, left panel). Coexpression were then subjected to immunoblot analysis with the mAb to human SIRP␣1 (upper panel). Duplicate precipitates were probed with the antibodies to NFB42 (lower panel). Brain extract was also subjected directly to immunoblot analysis to determine the total amounts of each protein. D, CHO-K1 cells were subjected to transient transfection with both 2 g of pTracerCMV encoding mouse SHPS-1 and 2 g of pcDNA3 encoding (or not) HA-tagged rat NFB42. Forty-eight hours after transfection, cell lysates were prepared and subjected to immunoprecipitation with mAb 12CA5 to the HA tag, and the resulting precipitates were subjected to immunoblot analysis with polyclonal antibodies to SHPS-1 (upper panel). Duplicate precipitates were probed with antibodies to NFB42 (lower panel). Cell lysates were also similarly probed to determine the expression level of each recombinant protein. Immature and mature forms of SHPS-1 are indicated. of wild-type NFB42 substantially reduced the binding to G. nivalis agglutinin of higher electrophoretic mobility forms of SHPS-1 (under reducing conditions) but increased that of slower migrating forms. In contrast, expression of NFB⌬P⌬F markedly enhanced the binding to G. nivalis agglutinin of high mobility forms of SHPS-1. These high mobility forms of SHPS-1 migrated as high molecular weight complexes under nonreducing conditions, suggesting that they represent misfolded SHPS-1 oligomerized by the formation of disulfide bonds (Fig.  5A, right panel). Expression of wild-type NFB42, but not that of NFB⌬P⌬F, substantially inhibited the formation of high molecular weight complexes of SHPS-1 induced by exposure of cells to thapsigargin (Fig. 5B), which blocks early processing of proteins in the ER. Given that terminal mannose residues are often exposed by folding intermediates of type I glycoproteins in the ER, these results suggest that SCF NFB42 interacts selectively with misfolded forms of SHPS-1 that are generated dur-ing protein maturation and thereby directs their elimination by the ERAD pathway.
Activation of SCF NFB42 Increases SHPS-1 Expression at the Cell Surface-We observed that the total amount of SHPS-1 in cells was substantially increased by expression of wild-type NFB42 (Figs. 1D and 5A). Of newly synthesized SHPS-1 molecules, only those that are folded into the correct tertiary structure in the ER are likely transported to the cell surface. To explore further the biological consequences of elimination of SHPS-1 by the ERAD pathway, we determined whether the presence of NFB42 affects SHPS-1 expression at the cell surface. Flow cytometry revealed that 5-10% of CHO cells subjected to transient transfection with SHPS-1 cDNA exhibited SHPS-1 immunoreactivity at the cell surface; this ratio was not substantially affected by cotransfection of the cells with wildtype NFB42 or NFB⌬P⌬F cDNAs (data not shown). However, coexpression of wild-type NFB42 markedly increased the pro-

FIG. 2. Structural requirements for the SHPS-1-NFB42 interaction.
A, brain extracts prepared from wild-type mice and from mice expressing a mutant form of SHPS-1 (SHPS-1-⌬cyto) that lacks most of the cytoplasmic region (cyto(Ϫ/Ϫ) mice) were subjected to immunoprecipitation (IP) with a mAb to mouse SHPS-1 or with control IgG. The resulting precipitates were then subjected to immunoblot analysis with antibodies to NFB42 (middle panel) or with a mAb to Skp1 (bottom panel). Duplicate precipitates were probed with polyclonal antibodies to SHPS-1 (top panel). Brain extracts were also subjected directly to immunoblot analysis with the same antibodies. B, CHO-K1 cells were transiently transfected with both 2 g of pcDNA3 encoding (or not) HA-tagged NFB42 and 2 g of pTracerCMV encoding either wild-type SHPS-1 (WT), an SHPS-1 mutant (4F) in which all four cytoplasmic tyrosine residues are replaced by phenylalanine, or an SHPS-1 mutant (JM) in which two juxtamembrane lysine residues are replaced by asparagine and glutamine, respectively. Cell lysates were prepared 48 h after transfection and were subjected to immunoprecipitation with a mAb to the HA tag. The resulting precipitates were subjected to immunoblot analysis with polyclonal antibodies to SHPS-1 (top panel). Duplicate precipitates were probed with antibodies to NFB42 (2nd panel). Cell lysates were also subjected directly to immunoblot analysis with the same antibodies (3rd and bottom panels). C, structures of HA-tagged wild-type (Full) and mutant NFB42 proteins. D, CHO-mSHPS-1 cells were transiently transfected with 2 g of pcDNA3 encoding HA-tagged wild-type or mutant NFB42 proteins (or with the empty vector), after which cell lysates were subjected to immunoprecipitation with the mAb to HA. The resulting precipitates were then subjected to immunoblot analysis with polyclonal antibodies to SHPS-1 (top panel) or with the mAb to Skp1 (middle panel). Duplicate precipitates were probed with antibodies to NFB42 (bottom panel). Cell lysates were also subjected directly to immunoblot analysis with the same antibodies. portion of cells with a relatively high level of SHPS-1 expression at the cell surface (Fig. 6A). Coexpression of NFB⌬P⌬F had no such effect. Coexpression of neither wild-type NFB42 nor NFB⌬P⌬F affected the amount of GFP synthesized as a tracer for the introduction of SHPS-1 cDNA (data not shown), suggesting that the effect of NFB42 on SHPS-1 expression was not due to variation in transfection efficiency or to a nonspecific action of the recombinant NFB42 protein.
Proteins misfolded in the ER generate a stress signal that increases the phosphorylation level of eIF2␣ and thereby inhibits protein synthesis (34,35). Overexpression of SHPS-1, but not that of some other transmembrane glycoproteins, in Melan-a cells increased the phosphorylation of eIF2␣ on Ser 51 (Fig. 6B; data not shown). This effect was prevented by coexpression of wild-type NFB42 but, in contrast, was enhanced by that of NFB⌬P⌬F (Fig. 6B). These results thus suggest that SCF NFB42 -mediated elimination of misfolded SHPS-1 results in up-regulation of the surface expression of the mature protein and that this effect may be attributable, at least in part, to dephosphorylation of eIF2␣. DISCUSSION We have shown that NFB42 and Skp1 interact with SHPS-1 both in mouse brain and in mouse melanocytes. NFB42 and Skp1 together with Cullin1 constitute the SCF NFB42 ubiquitin ligase complex, which we have now shown mediates the polyubiquitination of immature SHPS-1 and its subsequent degradation by the ERAD pathway, a process important for the elimination of misfolded proteins generated under conditions of ER stress. Forced expression of NFB42 resulted in an increase FIG. 3. SCF NFB42 ubiquitin ligase mediates polyubiquitination and proteasomal degradation of SHPS-1. A, CHO-K1 cells were transiently transfected with both 2 g of pcDNA3 encoding Myc epitope-tagged Cullin1 and 2 g of pcDNA3 encoding HA-tagged NFB42. Cell lysates were prepared 48 h after transfection and were subjected to immunoprecipitation (IP) with mAb 9E10 to the Myc tag or with control IgG. The resulting precipitates were subjected to immunoblot analysis with antibodies to NFB42 (lower panel). Duplicate precipitates were probed with polyclonal antibodies to Cullin1 (upper panel). Cell lysates were also subjected directly to immunoblot analysis with the same antibodies. B, Melan-a cells were transiently transfected with 3 g of pcDNA3 encoding (or not) Myc-tagged Skp1, after which cell lysates were subjected to immunoprecipitation with the mAb to Myc, and the immunoprecipitates were probed with antibodies to NFB42 (middle panel) or to Cullin1 (top panel). Duplicate precipitates were probed with a mAb to Skp1 (bottom panel). Cell lysates were also probed directly with the same antibodies. C, CHO-mSHPS-1 cells were transiently transfected with 2 g of pcDNA3 encoding (or not) HA-tagged NFB42. Twenty-four hours after transfection, the cells were incubated for an additional 10 h in the presence of 10 M MG132. Cell lysates were then prepared and subjected to immunoprecipitation with a mAb to mouse SHPS-1 or with control IgG, and the resulting precipitates were probed with a mAb to ubiquitin (top panel). Duplicate precipitates were probed with polyclonal antibodies to SHPS-1 (2nd panel). Cell lysates were also subjected directly to immunoblot analysis with antibodies to SHPS-1 (3rd panel) or to NFB42 (bottom panel). The position of polyubiquitinated SHPS-1 is indicated in the top panel. D, cell lysates prepared as in C were subjected to immunoprecipitation with a mAb to the HA tag or with control IgG, and the resulting precipitates were probed with polyclonal antibodies to SHPS-1 (upper panel). Duplicate precipitates were probed with antibodies to NFB42 (lower panel). Cell lysates were also probed directly with the same antibodies. SHPS-1 that associated with NFB42 in response to MG132 treatment is indicated by the asterisk. Immature and mature forms of SHPS-1 are indicated. in the amount of SHPS-1 at the cell surface, indicating that activation of SCF NFB42 promotes the biosynthesis of SHPS-1. During the course of this study, Yoshida et al. (36) showed that SCF NFB42 targets N-glycosylated proteins. These researchers identified integrin ␤ 1 as a potential target of NFB42 and also showed that expression of inactive NFB42 delayed elimination of typical ERAD substrates, although the biological significance of these observations was unclear. Our results now establish SHPS-1 as a physiological substrate of SCF NFB42 ; moreover, they uncover a previously unidentified posttranslational mechanism by which the elimination of misfolded protein mediated by this ubiquitin ligase relieves the ER stress response FIG. 6. Effects of overexpression of NFB42 on the expression of SHPS-1 at the cell surface. A, CHO-K1 cells were transiently transfected both with 2 g of pTracerCMV encoding SHPS-1 and with 2 g of pcDNA3 encoding (or not) wild-type NFB42 or NFB⌬P⌬F, as indicated. Thirty-six hours after transfection, the cells were detached from the culture dish and incubated with a mAb to mouse SHPS-1 or with control IgG (not shown). Immune complexes were then detected with Alexa Fluor 488-conjugated goat antibodies to rat IgG and flow cytometry. The percentages of SHPS-1-positive cells that exhibited strong (M1) or weak (M2) immunoreactivity are indicated. Data are representative of those obtained in three independent experiments. B, Melan-a cells were transiently transfected both with 3 g of pSR␣ encoding SHPS-1 and 2 g of pcDNA3 encoding HA-tagged NFB42, or with the corresponding empty vectors, as indicated. Cell lysates were prepared 36 h after transfection and were subjected to immunoblot analysis with antibodies to eIF2␣, to eIF2␣ phosphorylated on Ser 51 , and to NFB42 (upper panel). The amount of Ser 51 -phosphorylated eIF2␣ (eIF2␣-P) was quantified by scanning densitometry with the NIH Image program, normalized by the total amount of eIF2␣, and expressed as a percentage of the value for cells transfected with both empty vectors (lower panel). Data are representative of those obtained in three independent experiments. and supports the production of mature SHPS-1.
Several lines of evidence indicate that SCF NFB42 mediates elimination of SHPS-1 by the ERAD pathway. First, blockade of this pathway by the proteasomal inhibitor MG132 induced SHPS-1 to form aggresome-like structures in the cytoplasm, thus identifying SHPS-1 as an ERAD substrate. Second, NFB42 selectively associated with and mediated the polyubiquitination of immature forms of SHPS-1, and these events were enhanced by the inhibition of proteasome activity. Thus, SCF NFB42 likely acts as a bona fide E3 that targets immature SHPS-1 for proteasomal degradation. Third, expression of wildtype NFB42 suppressed, whereas that of the dominant negative mutant NFB⌬P⌬F enhanced, the formation of cellular aggregates containing SHPS-1 in response to inhibition of the proteasome. Together, these results are consistent with the notion that NFB42 binds to misfolded forms of SHPS-1 generated during protein maturation, and they suggest a role for SCF NFB42 in elimination of the misfolded protein by the ERAD pathway.
NFB42 and Skp1 are both cytosolic proteins and thus would be expected to be separated topologically from the region of SHPS-1 where they associate. NFB42 recognizes the N-linked high mannose oligosaccharides of proteins as they are translocated from the ER to the cytosol (36). SHPS-1 is differentially glycosylated in a tissue-specific manner; neuronal (brain-derived) SHPS-1, unlike the myeloid (spleen-derived) type, is almost completely devoid of galactose in its N-glycan structures, although both types of SHPS-1 possess a high content of oligomannosidic moieties (37). It therefore seems likely that the mannosylated extracellular region of SHPS-1, when dislocated from the ER membrane, directly binds NFB42.
Elimination of misfolded SHPS-1 by SCF NFB42 resulted in a marked increase in the expression of the mature protein at the cell surface. If the rate of protein synthesis exceeds the capacity of the ER to process the newly synthesized protein molecules, protein folding becomes compromised, and the deployment of incompletely folded or assembled proteins may result in detrimental effects. Eukaryotic cells thus coordinate protein folding in the ER with gene transcription and mRNA translation by a process known as the unfolded protein response (UPR) (38 -40). Furthermore, genetic evidence has revealed an essential regulatory link between ERAD and the UPR; dysfunction of the ubiquitin-proteasome system stabilizes ERAD substrates and results in persistent activation of the UPR (41,42). On the basis of these previous observations and our present data, we propose a role for the SCF NFB42 ubiquitin ligase in the regulation of SHPS-1 biosynthesis (Fig. 7). Functional activation of SCF NFB42 thus likely facilitates elimination of misfolded SHPS-1 that accumulates during cellular stress and thereby restores the protein folding capacity of the ER, resulting in attenuation of the UPR, de-repression of mRNA translation, and increased surface expression of SHPS-1. This hypothesis is supported by our observation that forced expression of NFB42 prevented the formation of SHPS-1 aggregates in response to thapsigargin-induced ER stress. Furthermore, NFB42 expression induced the dephosphorylation of eIF2␣, which is indicative of attenuation of the UPR, suggesting that eIF2␣ participates in the up-regulation of SHPS-1 expression. Given that incorrectly folded proteins might interfere with the transport of other normal proteins out of the ER (43), it is also possible that activation of SCF NFB42 facilitates SHPS-1 transport to the cell surface (Fig. 7). FIG. 7. Proposed role for the SCF NFB42 ubiquitin ligase in regulation of SHPS-1 biosynthesis. Misfolded forms of SHPS-1 generated constitutively or in response to ER stress are either retained in the ER, where they might associate with molecular chaperone such as calnexin, or translocated to the cytosol. In the absence of NFB42 (right), they form aggresomes in the cytosol and induce the phosphorylation of eIF2␣ through the action of an ER-resident stress-responsive protein kinase (or kinases), resulting in cytotoxicity and inhibition of SHPS-1 mRNA translation, respectively. The misfolded proteins might also inhibit SHPS-1 trafficking along the secretory pathway. In the presence of NFB42 (left), however, these effects are likely counteracted by SCF NFB42 -mediated elimination of the misfolded proteins by the ubiquitin-proteasome pathway. See text for further details. Note that intact SHPS-1 forms a cis dimer (H. Ohnishi and T. Matozaki, unpublished data).
The ERAD pathway protects cells against the pathological effects of protein aggregates (44), a role that is especially important in nonreplicating cells such as terminally differentiated neurons. Misfolding of the abundant glycoproteins of neuronal cells would thus be expected to exert toxic effects if not kept under control by this pathway. Given that terminal mannose residues are preferential targets of NFB42, it is possible that SCF NFB42 plays a general role in supporting the viability of adult neurons through the elimination of misfolded glycoproteins.